All-Cause Mortality and Bleeding Risks Associated With Serotonin Reuptake Inhibitors in Patients With End-Stage Kidney Disease and Depression: A Nationwide Cohort Study
Article information
Abstract
Objective
The prevalence of depression is high among patients with end stage kidney disease (ESKD). To date, there has been limited investigation into the comparative effects of antidepressant in patients with ESKD. This study aims to explore the association between type of antidepressant, incidence of all-cause death, and hospitalization for major bleeding in patients with ESKD and depression.
Methods
This study utilized data obtained from the Korean National Health Insurance Service Database. Patients with ESKD were divided into two groups: those prescribed strong serotonin reuptake inhibitors (SRIs) and those prescribed weak or intermediate SRIs.
Results
Over a mean follow-up of 2.46 years, the strong SRI group had a lower risk of all-cause death (hazard ratio [HR] 0.87, 95% confidence interval [CI] 0.81–0.93) and hospitalization for major bleeding (HR 0.84, 95% CI 0.79–0.90) with no increased risk of bleeding-related death (HR 1.05, 95% CI 0.80–1.37) compared to the weak or intermediate SRI group. The protective effects of strong SRI use for all-cause death and hospitalization for major bleeding remained consistent in those prescribed SSRIs for less than 120 days (death: HR 0.85, 95% CI 0.80–0.92; hospitalization for major bleeding: HR 0.84, 95% CI 0.78–0.90), and in patients aged below 75 years (death: HR 0.83, 95% CI 0.76–0.90; hospitalization for major bleeding: HR 0.81, 95% CI 0.75–0.87).
Conclusion
In patients with ESKD and depression, the use of strong SRIs was associated with a reduced risk of all-cause death and major bleeding hospitalization compared to the use of weak or intermediate SRIs.
INTRODUCTION
Cardiovascular complications are a significant mortality risk in patients with chronic kidney disease (CKD) [1]. Common risk factors such as bleeding tendencies, albuminuria, anemia, malnutrition, vitamin D deficiency, secondary hyperparathyroidism, and fluid overload, contribute to a markedly reduced quality of life and poor prognosis for CKD patients [2,3]. Among those receiving hemodialysis, intracranial hemorrhage and gastrointestinal (GI) bleeding occur frequently, and these bleeding risks are associated with a higher likelihood of mortality [4,5]. Consequently, the elevated risk of bleeding complicates the prescription of anticoagulants or antiplatelet agents for hemodialysis patients, compelling nephrologists to vigilantly balance the risk of thrombosis against that of hemorrhage.
Mental health disorders, including depression, anxiety, stress reactions, adjustment disorders, somatoform/conversion disorders, and substance use disorders, are also prevalent among patients with end-stage kidney disease (ESKD) [6,7]. Approximately 20% to 40% of ESKD patients suffer from major depressive disorder, which is linked to higher mortality rates and diminished quality of life [8-12]. Although depression is a frequent and serious condition in ESKD patients undergoing intensive dialysis treatment, evidence supporting the efficacy of pharmacotherapy and psychological interventions remains limited.
Nonetheless, serotonin reuptake inhibitors (SRIs) are commonly prescribed as a first-line treatment for depression by nephrologists and psychiatrists. Observational cohort studies in the general population have suggested that bleeding risks increase with higher doses of antidepressants [13-16]. Given the significantly elevated rates of mortality, bleeding complications, and depression among ESKD patients, this study aims to assess, first, the risk of all-cause mortality and, second, the risk of major bleeding in relation to SRI dosage levels and depression severity in this population.
METHODS
Data sources
Data for this study was extracted from the National Health Insurance Service (NHIS) database which is operated by the Korean government. The NHIS covers 97% of all Korean people (almost 50 million citizens), and includes data from hospitals documenting information about inpatient and outpatient visitations, procedures, prescriptions and national health examination data. This database includes drug prescription codes, disease-specific codes (for national insurance discounts), diagnostic codes (according to the International Classification of Diseases [ICD] 10th Edition), in- and outpatient hospital visitation records, procedure codes, and operation codes. These data resources are widely validated and used for epidemiologic studies [17]. The Statistics Korea database, which provides information on cause of death and date of death (http://mdis.kostat.go.kr), was merged with the NHIS database. The data are available with the approval of the NHIS (NHIS-2021-1-243) through the Korean National Health Insurance Sharing Service (http://nhiss.nhis.or.kr). The codes used to define each diagnosis and drug (diagnostic code, prescription code, disease-specific codes, or medications) are shown Supplementary Tables 1 and 2. The current study was reviewed and approved by the Institutional Review Board of Wonju Severance Christian Hospital (IRB No. CR323324). Informed consent was waived because anonymous and deidentified information was used for analyses.
Study population
This study included patients who were diagnosed with ESKD who received the relevant prescription code between January 2002 and December 2019 (n=522,794). We excluded patients who were not prescribed antidepressants (n=270,740), patients who were prescribed antidepressant medication before starting renal replacement therapy (n=159,005), patients who had invalid data (n=4,870), patients who experienced bleeding before receiving an antidepressant prescription (n=35,027), and patients who were prescribed more than one type of antidepressant (n=15,873) (Figure 1). We divided the remaining 37,279 ESKD patients into two groups: those who were prescribed strong SRIs (strong group; n=5,307) and those who were prescribed weak or intermediate SRIs (weak or intermediate group; n=31,972). We evaluated whether the risk of bleeding varied according to the degree of serotonin reuptake inhibition of an antidepressant (strong and intermediate or weak).
Flow diagram of the study population. ESKD, end-stage kidney disease; RRT, renal replacement therapy; NSAID, non-steroidal anti-inflammatory drug.
In this study the strong SRIs were selective serotonin reuptake inhibitors (SSRI) and serotonin-norepinephrine reuptake inhibitors (SNRI), while the weak or intermediate SRIs were noradrenergic and specific serotonergic antidepressant (NaSSA), tricyclic antidepressant (TCA), monoamine oxidase inhibitors (MAOi), and norepinephrine reuptake inhibitor (NRI). This classification is based on the affinity of each antidepressant for the serotonin transporter: agents with higher transporter affinity exhibit greater inhibition of serotonin reuptake and more pronounced antiplatelet effects [18].
Propensity score matching
Propensity score matching (PSM) was performed to control for confounding by ensuring comparability between the strong and weak or intermediate antidepressant groups. The strong and weak or intermediate groups were matched at a 1:3 ratio based on propensity scores using greedy matching techniques with a caliper of 0.2 standard deviations. Variables relevant to treatment of depression and ESKD patient health such as age, sex, Charlson Comorbidity Index (CCI) score, duration from dialysis to antidepressant prescription, and medication use (antiplatelet agents, anticoagulants, and nonsteroidal anti-inflammatory drugs [NSAIDs]) were used to generate the propensity score. Table 1 shows the covariate balance between the groups with standardized mean differences before and after matching.
Baseline characteristics of patients
Outcome measures and follow-up
The primary outcome was all-cause mortality. The secondary outcomes were the occurrence of hospitalization for major bleeding, including GI bleeding and intracranial hemorrhage. Past history of underlying disease such as diabetes mellitus, hypertension, cardiovascular disease, cerebrovascular disease, and ESKD were confirmed by the assignment of at least two claims within 1 year using the ICD-10 code or medications (Supplementary Tables 1 and 2). The region of residence was classified as rural (Gyeonggi, Gangwon, Gyeongsangbuk, Gyeongsangnam, Chungcheongbuk, Chungcheongnam, Jeollabuk, Jeollanam, and Jeju) and urban (Seoul, Sejong, Busan, Incheon, Daegu, Gwangju, Daejeon, and Ulsan) [19]. The CCI score was calculated from the ICD-10 codes [11]. GI bleeding and intracranial bleeding were confirmed by the ICD-10 code [19]. Hospitalization for major bleeding was defined as a diagnosis of GI bleeding or hemorrhagic stroke requiring hospital admission. We evaluated whether the risk of bleeding varied according to the degree of SRI by antidepressant (classified as strong and intermediate or weak), with all antidepressants were reclassified accordingly (Supplementary Table 2) [14]. The study population was followed until the date of mortality or December 31, 2019, whichever occurred first.
Statistical analysis
Baseline characteristics were compared between the two groups before and after matching using the t-test or chi-square test, as appropriate. Continuous variables were expressed using the mean and standard deviation. Categorical variables were expressed with numbers and percentages. Kaplan–Meier survival curves and the log-rank test were used to compare cumulative outcome incidences. Hazard ratios (HRs) were obtained for each outcome before and after PSM using Cox proportional hazard regression analysis. Subgroup analyses were performed based on the duration of prescription days (<120 days or ≥120 days), concurrent using or not using bleeding related medication (antiplatelets or anticoagulants), and aged above 75 years or blow 75 years. The age categorization aligns with previous studies that have demonstrated differences in CKD and ESKD progression and mortality outcomes among patients aged 75 years or older [20,21]. All p values were two-sided, and those less than 0.05 were considered significant. The analyses were performed using the statistical package SAS 9.4 (SAS Institute) and R version 4.20 for Windows (https://cran.r-project.org/).
RESULTS
Patient characteristics
Baseline characteristics of the strong and weak or intermediate groups before and after matching are shown in Table 1. Detail list of the drugs included in the two groups were shown Supplementary Tables 2 and 3. The total antidepressant prescription days for the strong group and weak or intermediate group were 259.0 (±573.8) days and 494.5 (±888.6) days, respectively (Supplementary Table 4). Compared with the strong group, more patients in the weak or intermediate group used bleeding-related drugs, such as anticoagulants, NSAID, and antiplatelet drugs. In PSM, 5,297 patients in the strong group were matched to 15,891 patients in the weak or intermediate group.
Clinical outcomes of ESKD patients prescribed antidepressants
During the mean follow-up period of 2.46 years among propensity score–matched patients, 1,437 (27.1%) patients in the strong group and 5,694 (35.8%) patients in the weak or intermediate group died. The most common cause of death was unknown in both groups (21.6% in the strong group and 25.7% in the weak or intermediate group) (Supplementary Table 5). During the mean follow-up period of 2.8 years, 1,432 (27.0%) patients in the strong group and 5,626 (35.4%) patients in the weak or intermediate group were hospitalized for major bleeding. The 1-, 3-year cumulative incidence of all-cause death were 13.0% and 20.6% in the strong group and 13.6% and 23.3% in the weak or intermediate group. The 1-, 3-year cumulative incidence of hospitalization for major bleeding were 14.5% and 21.9% in the strong group and 16.7% and 26.1% in the weak or intermediate group. Details of the outcome frequencies are shown Supplementary Tables 5 and 6.
After PSM, Cox regression analysis showed that the strong group was associated with a lower risk of all-cause death (HR 0.87; 95% confidence interval [CI] 0.81–0.93), and hospitalization for major bleeding (HR 0.84; 95% CI 0.79–0.90), any bleeding (HR 0.87; 95% CI 0.82–0.92), and GI bleeding (HR 0.85; 95% CI 0.79–0.91) compared with the weak or intermediate group. However, the strong group was not associated with a reduced risk of bleeding-related death (HR 1.05; 95% CI 0.80–1.37) and intracranial bleeding (HR 0.83; 95% CI 0.66–1.04) compared with the weak or intermediate group (Table 2). The Kaplan–Meier curves for all-cause death and hospitalization for major bleeding, GI bleeding, intracranial hemorrhage and other outcomes are similar and shown in Figure 2 and Supplementary Figure 1.
Comparison of clinical outcomes between the strong group and weak or intermediate group
Kaplan–Meier curve analysis for cumulative incidence of all-cause mortality (A), hospitalization for major bleeding (B), intracranial hemorrhage (C), gastrointestinal bleeding (D).
The subgroup analysis showed that compared with the weak or intermediate group, the strong group was associated with a lower risk of all-cause death, hospitalization for major bleeding, any bleeding, and GI bleeding even when antidepressants were prescribed for <120 days; concurrent not use of bleeding-related drugs (antiplatelet agents or anticoagulants), and aged under 75 years (Table 3). However, those beneficial effects were not prominent when antidepressants were compared between SSRIs and non-SSRIs (Supplementary Table 7).
Comparison of clinical outcomes between the strong SSRI group and intermediate or weak SSRI group (according to prescription days)
DISCUSSION
The study cohort consisted of 252,054 of 522,794 ESKD patients (48.2%) who received prescription code and who were prescribed antidepressants regardless of past history of depression. Among them, 88,179 (35.0%) patients were prescribed antidepressants after starting renal replacement therapy. Most ESKD patients (85.8%) were prescribed weak or intermediate SRIs. Compared with the weak or intermediate group, the strong group was associated with a lower risk of all-cause death and hospitalization for major bleeding. Those associations were consistent even in patients who were prescribed an SRIs for fewer than 120 days, no concurrent prescribed bleeding related drugs, and in those aged below 75 years old.
The pathophysiology of uremic bleeding is multifaceted. In uremic status, von Willebrand factor activity is decreased, while prostaglandin I2, cyclic GMP, nitric oxide, and uremic toxins are elevated [22,23]. These changes impair platelet and vessel wall interactions, adhesion, platelet secretion, and aggregation, leading to an increased bleeding tendency [24]. Although dialysis removes the uremic toxins, associated with platelet dysfunction, the bleeding tendency persists in ESKD patients because it does not remove them completely or efficiently. Furthermore, anemia aggravates platelet dysfunction [24,25], and the prevalence of antithrombotic drug use is higher in ESKD patients than in the general population [26]. Patients need to use anticoagulants during hemodialysis sessions to prevent clotting in the extracorporeal system. Compared with the general population, hemodialysis and peritoneal dialysis patients have shown an increased risk of bleeding amounting to 34.6–60.8/1,000 person-years [5].
Nearly all serotonin in the human body is stored in platelets. Following platelet activation and vascular injury, serotonin is released to promote hemostasis [27]. Since platelets cannot synthesize serotonin, inhibition of serotonin transport leads to a reduction in serotonin concentrations within platelets, potentially increasing the risk of abnormal bleeding [28]. Not only SSRIs, but also other antidepressants inhibit serotonin reuptake in the platelets, thereby inhibiting platelet aggregation and platelet thrombus formation [29,30]. Several cohort studies and meta-analysis were conducted to analyze the risk of bleeding associated with antidepressant prescription [13-16,31].
To the best of our knowledge, this is the first population-based cohort study to not only assess all-cause mortality but to also compare bleeding complications based on the degree of serotonin reuptake inhibition in patients with ESKD. Some studies have investigated the effects of antidepressants on dialysis patients; however, these studies either did not report bleeding-related side effects or reported no bleeding-related events in a small patient population [29-35]. For ESKD patients, the action sites of antidepressants especially the platelets are damaged by decreased renal function before antidepressants use, which may explain why strong serotonin reuptake inhibition reduce bleeding. More research is needed to evaluate the exact mechanism.
Although some observational studies have shown an association between SSRIs and an increased risk of bleeding compared to non-SSRI antidepressants, other studies have suggested that non-SSRI antidepressants may also elevate the risk of bleeding [36-40]. Furthermore, meta-analysis of randomized controlled trials reported that compared to placebo SSRIs did not increase risk of bleeding [41]. Based on our findings, it appears that the degree of serotonin reuptake inhibition, rather than the classification of SSRI versus non-SSRI, influences the bleeding tendency. Studies on antidepressant-anticoagulant interactions have shown that both SSRIs and non-SSRIs like amitriptyline and trazodone can affect bleeding, with effects varying depending on their interactions with other drugs [42-44]. ESKD patients, who are often exposed to multiple medications and have unique factors such as drug interactions and platelet dysfunction, may experience different mechanisms compared to the general population. Therefore, further research is essential to explore these differences.
A Dutch nationwide cohort study showed that dialysis patients are prescribed antidepressants more frequently than the general population and that SSRIs were the most prescribed antidepressant, followed by TCAs [45]. The US Renal Data System has reported that 21.5% of hemodialysis patients and 19.5% of peritoneal dialysis patients are prescribed SSRIs, which are the 13th and 14th most commonly prescribed medications among ESKD patients in the US [46]. Some guidelines recommend SSRIs as first-line antidepressants for CKD patients [47,48].
In addition to treating depression, non-SSRIs are used for various medical purposes in ESKD patients. Many experience chronic pain due to conditions such as diabetic neuropathy, peripheral vascular disease, uremic neuropathy, muscle cramps, and dialysis-related pain. For neuropathic pain, drugs like amitriptyline (TCA) and SNRIs are commonly prescribed [49,50]. Additionally, dialysis patients often suffer from reduced sleep quality due to conditions like primary insomnia, restless legs syndrome, and obstructive sleep apnea. Medications such as amitriptyline and trazodone are frequently used to address these issues [51-55]. Given our study’s findings, it may be worthwhile to consider using strong SRIs instead of weak or intermediate SRIs for antidepressant treatment in ESKD patients.
These findings carry important implications for clinical decision- making in ESKD. While conventional guidance often classifies antidepressants broadly as SSRIs versus non-SSRIs, our results suggest that such classifications may be insufficient in predicting safety outcomes. Instead, the degree of serotonin reuptake inhibition may offer a more precise and clinically relevant framework. In particular, strong SRIs were not associated with increased bleeding or mortality risks, in contrast to prior assumptions, and may represent a safer therapeutic option in this vulnerable population. Incorporating pharmacologic potency profiles into future guidelines may help optimize antidepressant selection for patients with ESKD.
This study has several limitations. The NHIS database used in this study did not allow us to fully eliminate selection bias, despite PSM. For instance, we lacked information from questionnaires, laboratory results, and specific details on depression severity, which may have influenced the outcomes. Moreover, the limited clinical information posed challenges in accurately calculating CCI scores. Likewise, the classification of residential areas into rural and urban was based on a rough distinction, which may not fully capture regional differences. Additionally, certain patient characteristics, such as anxiety, stress, and alcohol use, could not be identified, as they were not included in the available diagnostic codes. As a result, we were unable to determine the most frequent cause of death in the cohort, which may have included cardiovascular death.
The relatively small number of patients using strong SRIs also limits the generalizability of the comparison between strong SRIs and weak or intermediate SRIs. Further studies with larger and more diverse datasets are needed to better estimate this association. The study participants were all from Korea, so the findings may not be generalizable to other populations. Furthermore, pharmacoepidemiology research using secondary analyses from large databases inherently cannot determine whether participants adhered to the antidepressant prescriptions, which is another limitation of the study. The mean follow-up period among propensity-matched patients may not have been long enough for the outcomes to manifest, potentially leading to immortal time bias. Additionally, the study lacked data on the adequacy of dialysis delivery, which is thought to affect bleeding risk.
Although antidepressants may influence clinical outcomes in ESKD patients, we are not claiming that one antidepressant is more effective than another in treating depression in this cohort. The primary outcome of this study was all-cause mortality, and while SSRIs have been shown to reduce mortality, we cannot infer a direct or causal relationship between antidepressant use and the bleeding complications observed in this study. Our results do not establish causality, but the observed association between strong SRIs and a reduced risk of bleeding complications suggests that the degree of serotonin reuptake inhibition could influence bleeding risks in ESKD patients.
Despite the study’s limitations, the results of this nationwide, population-based cohort study suggest that, among patients with ESKD and depression, the use of strong SRIs was associated with lower all-cause mortality and hospitalization for major bleeding compared to the use of weak or intermediate SRIs. These findings suggest that strong SRIs may be a safer option for ESKD patients who are prescribed antidepressants, unless there is a specific contraindication.
Supplementary Materials
The Supplement is available with this article at https://doi.org/10.30773/pi.2024.0359.
Diagnosis, procedure, and specific codes
Drug codes used for the National Health Insurance Korea database analysis
Detail list of the drugs included in the two groups
Antidepressants prescription duration
Details of outcome frequency (matched analysis)
Details of major outcome frequency (matched analysis)
Comparison of clinical outcomes between the SSRI group and non-SSRI group
Kaplan–Meier curve for each outcome. A: Bleeding death. B: All bleeding. C: Urinary tract bleeding. D: Airway bleeding. E: Upper gastrointestinal bleeding. F: Lower gastrointestinal bleeding. G: Ischemic gastrointestinal bleeding. H: Other bleeding.
Notes
Availability of Data and Material
The data used and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Conflicts of Interest
The authors have no potential conflicts of interest to disclose.
Author Contributions
Conceptualization: Jinhee Lee, Jun Young Lee. Data curation: Jinhee Lee, Minae Park, Jun Young Lee. Formal analysis: Minae Park, Jun Young Lee. Funding acquisition: Jun Young Lee. Investigation: Jinhee Lee, Minae Park, Sojeong Park, Yun Seob Jung, Keunryul Park, Curmairah Keisha Cambridge, Sejeong Park, Byoung Geun Han, Jun Young Lee. Methodology: Jinhee Lee, Minae Park, Sojeong Park, Yun Seob Jung, Keunryul Park, Curmairah Keisha Cambridge, Sejeong Park, Byoung Geun Han, Jun Young Lee, Dong Hui Shin. Resources: Jinhee Lee, Minae Park, Sojeong Park, Yun Seob Jung, Keunryul Park, Curmairah Keisha Cambridge, Sejeong Park, Byoung Geun Han, Jun Young Lee. Software: Jinhee Lee, Minae Park, Sojeong Park, Yun Seob Jung, Keunryul Park, Curmairah Keisha Cambridge, Sejeong Park, Byoung Geun Han, Jun Young Lee. Supervision: Jinhee Lee, Jun Young Lee. Validation: Jinhee Lee, Jun Young Lee. Visualization: Jinhee Lee, Jun Young Lee, Dong Hui Shin. Writing—original draft: Jinhee Lee, Jun Young Lee, Dong Hui Shin. Writing—review & editing: Jinhee Lee, Jun Young Lee.
Funding Statement
This work was supported by the Government of the Republic of Korea (Ministry of Science and ICT, MSIT) and the National Research Foundation of Korea (NRF) under grant number (NRF-RS-2024-00466764, FY2025). Additionally, support was provided by the NRF of Korea, including the Medical Research Center Program (RS-2024-00409403). This research was also supported by the Basic Science Research Program through the NRF, funded by the Ministry of Education (RS-2023-00240674).
This work was also supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (RS-2024-00439231 and RS-2024-00439956). This research was also supported by the Bio&Medical Technology Development Program of the NRF funded by the Korean government (MSIT) (No. RS-2024-00440802). This research was supported by a grant of the Medical data-driven hospital support project through the Korea Health Information Service (KHIS), funded by the Ministry of Health & Welfare, Republic of Korea.
Acknowledgments
None